Work Method Foundry Field Notes About Edge Brief
← All work

Selected work · prediction & hedging · venture

Capitalizing on poor weather.

An advanced weather-prediction and hedging terminal, built for distribution businesses whose revenue moves with the temperature.

Weather Terminal DeskContractsFleetHedge Book LIVE · SEASON 2025–26
Season HDD vs normal
−8.1%
10-day forecast HDD
312
Hedge coverage
68%
Hedge book P&L
+$412K
Projected volume
31.4M gal
Cumulative season HDD Actual Normal Forecast band
NOVDECJAN FEBMAR
JAN 14 · HDD 2,214 vs normal 2,410
volume gap −1.9M gal · hedge payout +$188K
2,214
HedgeUnwind
ContractHDD-CHI-JAN
Last528.65
StructurePut · 2,300 strike
Coverage of expected volume
0%25%50%● 68%100%
Warm-winter scenario · −10% HDD
Volume loss−$1.9M margin
Hedge payout+$1.6M
Net exposure−$0.3M
Depot · Northfield14 routes
$0.142
cost / gal7-day avg97.2%on-time in fullsurge +2drivers thu
Depot · Ridgeway11 routes
$0.156
cost / gal7-day avg95.8%on-time in fullnormalstaffing
Depot · Lake Mills9 routes
$0.149
cost / gal7-day avg96.4%on-time in full512 tanks<30% · queued
HDD-NYC-DEC487.30▲ 3.4 CDD-DFW-JUL612.85▼ 1.8 WX-MID-Q11041.20▲ 0.9 HDD-CHI-JAN528.65▲ 5.1 FILL · BUY 40 HDD-CHI-JAN@528.65filled NYC38.4°▼ 2.1 CHI27.8°▼ 3.6 DFW61.2°▲ 4.4 HDD-NYC-DEC487.30▲ 3.4 CDD-DFW-JUL612.85▼ 1.8 WX-MID-Q11041.20▲ 0.9 HDD-CHI-JAN528.65▲ 5.1

The engagement

A middle-market fuel distributor asked us to look into something its ownership had treated as a fact of life for three generations: the weather was deciding their year. Warm winters quietly removed ten to twelve percent of heating-season volume, cold snaps blew up delivery budgets, and neither exposure appeared anywhere in the company’s planning except as an apology in the annual letter.

The engagement began the way all of ours do — with an Edge Brief, followed by a secure data room. Our agents read several years of the company’s operating and financial data against external market data, and surfaced a ranked ledger of problems worth money. Weather appeared near the top twice: once as missing revenue, once as inflated delivery cost. That distinction turned out to be the whole engagement.

The process

Discover → Architect → Build → Commercialize.

Discover
The data room did what data rooms are for: it separated what ownership believed from what the numbers showed. The two weather exposures behaved nothing alike — lost volume tracked a listed temperature index closely enough to hedge, while delivery-cost overruns tracked fuel, labor, and delay, which no index reaches.
Architect
We generated solution candidates deliberately wide — from pricing changes to fleet contracts to financial instruments — and tested each against the opportunity matrix. Most died there, as most should. Two survived: hedge the demand exposure with listed degree-day structures, and convert the same forecast into dispatch decisions for the cost exposure.
Build
The two survivors were built as one working environment and tested against prior seasons before any position was ever considered. What held up in the backtests is what you see on this page; what didn’t, we discarded.
Commercialize
The system went into use — and the screening that starts every engagement showed this problem was not one company’s problem. It belongs to an entire class of temperature-driven businesses, which is why the solution now operates as Weather Terminal, a Modven venture. One problem, paid twice, exactly as the Foundry intends.

What the process selected · 01

HEDGE TICKET · HDD-CHI-JAN
StructurePut · 2,300 strike
Last528.65 ▲ 5.1
Coverage of expected volume
0%25%50%● 68%100%
Warm-winter scenario · −10% HDD
Volume loss−$1.9M margin
Hedge payout+$1.6M
Net exposure−$0.3M

The hedge, written in the owner’s language.

This candidate won selection because it cleared the bar the others couldn’t: the decision-maker had to be able to defend it to their own board. Position size reads as coverage of expected volume, the downside case is stated in margin dollars, and the scenario math sits above the button. Instruments like these have existed for decades; what was missing was a version a mid-market owner could understand and act on without a trading desk.

What the process selected · 02

The forecast, converted into dispatch.

The second survivor answered the exposure the first one couldn’t touch. Testing showed delivery-cost overruns correlate too weakly with any tradable index to hedge economically — but they respond immediately to earlier, better dispatch. So the same model that prices the hedge drives the operating board: surge staffing flagged days ahead, low tanks queued before the cold arrives, delivered cost tracked against the weather that produced it. One model, both halves of the P&L, each treated the way the data said it should be.

10-DAY OPS BOARD · ALL DEPOTS
WEDTHUFRISAT–SUN COLD SNAPMON
NorthfieldSurge +2 drivers · Thu
Lake Mills512 tanks <30% · queue now
RidgewayNormal · 11 routes

Where it stands

In production, and now a Modven venture. The engagement solved one company’s exposure; the screen showed the problem belonged to an industry, and Weather Terminal is the second harvest. The limits travel with it: every hedge carries basis risk between a station index and one company’s micro-climate, so coverage is sized to measured sensitivity — never speculation dressed as protection.

Next case: Anvil →

Your business is holding problems worth money. Let’s find them.

Every engagement begins the same way: one paid conversation, one written brief, one clear answer about the edge worth building.

See more work